Our previous research examined the effectiveness of cognitive training in nondemented elderly subjects with memory complaints. We have studied the comparative effectiveness of different "pretraining" components, such as training in the use of visual imagery or verbal elaboration, in enhancing older peoples' ability to benefit from memory training using associative (mnemonic) devices. These programs have been successful overall, yet there remains a large variability of individual response across different ages, abilities and types of training. In our latest work we found an interaction between age and difficulty of training task. The general aim of the current proposal is to increase our understanding of this interaction with the goal of modifying our training programs to enhance their effectiveness for old-old individuals teaming relatively challenging mnemonic devices. Our proposed study will manipulate experimentally the type of pretraining and amount of training time given to subjects to learn a complex mnemonic. To identify factors that predict response to treatment, all subjects will receive a cognitive evaluation prior to starting the training program. Our hypotheses are: Hypothesis 1: Pretraining. A comprehensive pretraining program will enhance the ability of old-old subjects to profit from complex mnemonic training when compared to an active control pretraining and a placebo control pretraining. Hypothesis 2: Training Time. Increased training time to learn a complex mnemonic technique will enhance the ability of the old-old subjects to profit from training. Increased training time when added to a comprehensive pretraining program will have better results than either the comprehensive pretraining or additional learning time alone. Hypothesis 3: Predictors of Response to Training. Individual differences in success of training will vary according to performance on tests of cognitive processing. Furthermore, subjects who carry the diagnosis of Age Associated Memory Impairment (AAMI) will not respond as well to interventions as do subjects without this diagnosis (non-AAMI). Our hypotheses will be tested in one large comparative group design study. Each of five Conditions will have 100 completed subjects consisting of young-old with and without AAMI and old-old with and without AAMI. All subjects will be screened for AAMI and receive an evaluation of selected cognitive functions before entering the study.